Errors in Measurement

UNIT 2 ERRORS IN MEASUREMENT

Structure
2.1 Introduction
Objectives

2.2 Classification of Errors

2.2.1 Gross Errors
2.2.2 Systematic Errors
2.2.3 Random Errors

2.3 Accuracy and Precision

2.4 Calibration of the Instrument
2.5 Analysis of the Errors
2.5.1 Error Analysis on Common Sense Basis
2.5.2 Uncertainty Analysis
2.5.3 Statistical Analysis of Experimental Data
2.5.4 Method of Least Squares
2.5.5 Analysis of the Probability of Errors
2.5.6 Analysis of Limiting Errors

2.6 Summary
2.7 Key Words
2.8 Answers to SAQs

2.1 INTRODUCTION
The measurement of a quantity is based on some International fundamental standards.
These fundamental standards are perfectly accurate, while others are derived from these.
These derived standards are not perfectly accurate in spite of all precautions. In general,
measurement of any quantity is done by comparing with derived standards which
themselves are not perfectly accurate. So, the error in the measurement is not only due to
error in methods but also due to standards (derived) not being perfectly accurate. Thus,
the measurement with 100% accuracy is not possible with any method.
Error in the measurement of a physical quantity is its deviation from actual value. If an
experimenter knew the error, he or she would correct it and it would no longer be an
error. In other words, the real errors in experimental data are those factors that are always
vague to some extent and carry some amount of uncertainty. A reasonable definition of
experimental uncertainty may be taken as the possible value the error may have. The
uncertainty may vary a great deal depending upon the circumstances of the experiment.
Perhaps it is better to speak of experimental uncertainty instead of experimental error
because the magnitude of an error is uncertain.
At this point, we may mention some of the types of errors that cause uncertainty is an
experimental in measurement. First, there can always be those gross blunders in
apparatus or instrument construction which may invalidate the data. Second, there may
be certain fixed errors which will cause repeated readings to be in error by roughly some
amount but for some unknown reasons. These are sometimes called systematic errors.
Third, there are the random errors, which may be caused by personal fluctuation, random
electronic fluctuation in apparatus or instruments, various influences of friction, etc.
Objectives
After studying this unit, you should be able to
• understand the nature of errors and their sources in the measurement,
19
Principle of • know accuracy and precision in the measurement, and
Measurementgs and
Measuring Techniques • explain the various methods of analysis of the errors.

2.2 CLASSIFICATION OF ERRORS

Errors will creep into all measurement regardless of the care which is exerted. But it is
important for the person performing the experiment to take proper care so that the error
can be minimized. Some of the errors are of random in nature, some will be due to gross
blunder on the part of the experimenter and other will be due to the unknown reasons
which are constant in nature.
Thus, we see that there are different sources of errors and generally errors are classified
mainly into three categories as follows:
(a) Gross errors
(b) Systematic errors
(c) Random errors
2.2.1 Gross Errors
These errors are due to the gross blunder on the part of the experimenters or observers.
These errors are caused by mistake in using instruments, recording data and calculating
measurement results. For example: A person may read a pressure gage indicating
1.01 N/m2 as 1.10 N/m2. Someone may have a bad habit of memorizing data at a time of
reading and writing a number of data together at later time. This may cause error in the
data. Errors may be made in calculating the final results. Another gross error arises when
an experimenter makes use (by mistake) of an ordinary flow meter having poor
sensitivity to measure low pressure in a system.
2.2.2 Systematic Errors
These are inherent errors of apparatus or method. These errors always give a constant
deviation. On the basis of the sources of errors, systematic errors may be divided into
following sub-categories :
Constructional Error
None of the apparatus can be constructed to satisfy all specifications completely.
This is the reason of giving guarantee within a limit. Therefore, a manufacturers
always mention the minimum possible errors in the construction of the
instruments.
Errors in Reading or Observation
Following are some of the reasons of errors in results of the indicating
instruments :
(a) Construction of the Scale : There is a possibility of error due to the
division of the scale not being uniform and clear.
(b) Fitness and Straightness of the Pointer : If the pointer is not fine
and straight, then it always gives the error in the reading.
(c) Parallax : Without a mirror under the pointer there may be parallax
error in reading.
(d) Efficiency or Skillness of the Observer : Error in the reading is
largely dependent upon the skillness of the observer by which reading
is noted accurately.
Determination Error
It is due to the indefiniteness in final adjustment of measuring apparatus. For
example, Maxwell Bridge method of measuring inductances, it is difficult to find the
differences in sound of head phones for small change in resistance at the time of final
20
adjustment. The error varies from person to person.
Natural Error of the Method Errors in Measurement

No method is perfect. Each method has some errors. Of course, error varies from
one method to another. The following example is given to clear the point. The two
systems (Figure 2.1) of Spring-mass-damper are given below where the
components (mass, damper and spring) are connected in different manner. In both
the systems, the purpose is to study the response of the systems for a forced
excitation. Both methods will give error for displacement measurement by virtue
of connection.
F F

x x

B M M
K

K B

Figure 2.1 : Two Ways of Connections of M, B and K

Error due to Other Factors

(a) Temperature Variation : Variation in temperature not only changes the
values of the parameters but also brings changes in the reading of the
instrument. For a consistent error, the temperature must be constant.
(b) Effect of the Time on Instruments : There is a possibility of change in
calibration error in the instrument with time. This may be called ageing of
the instrument.
(c) Effect of External Electrostatic and Magnetic Fields : These electrostatic
and magnetic fields influence the readings of instruments. These effects can
be minimized by proper shielding.
(d) Mechanical Error : Friction between stationary and rotating parts and
residual torsion in suspension wire cause errors in instruments. So, checking
should be applied. Generally, these errors may be checked from time to
time.
2.2.3 Random Errors
After corrections have been applied for all the parameters whose influences are known,
there is left a residue of deviation. These are random error and their magnitudes are not
constant. Persons performing the experiment have no control over the origin of these
errors. These errors are due to so many reasons such as noise and fatigue in the working
persons. These errors may be either positive or negative. To these errors the law of
probability may be applied. Generally, these errors may be minimized by taking average
of a large number of readings.
SAQ 1
(a) What is the difference between error and accuracy?
(b) What do you mean by uncertainty in measurement?
(c) What is the difference between fixed error and random error?
(d) Mention the role of the experimenter to minimize error in measurement.
(e) Identify the nature of error in the following cases :

21
Principle of (i) The magnitude of a known voltage source of 100 V was measured
Measurementgs and with a voltmeter. Five readings were taken. The indicating readings
Measuring Techniques
were 101, 100, 102, 100 and 99.
(ii) The temperature of a hot fluid is 200oC. A glass bulb thermometer is
used to measure the same for five times. The temperature indicated by
the thermometer in each case is 180oC.
(iii) Five students were asked to take the readings of a pressure gage. The
readings noted by them were 1.5 N/m2, 1.51 N/m2, 1.49 N/m2,
1.48 N/m2 and 1.5 N/m2.
(iv) Due to fluctuation of the voltage source, the pointer of the voltmeter
indicates maximum and minimum readings of 230 and 220 volts
respectively but the reading taken by the experimenter is 203 V.

2.3 ACCURACY AND PRECISION

Accuracy plays an important role in the measurement of any quantity. The word
‘precision’ is often used in place of accuracy as if they are interchangeable. The accuracy
of measurement is defined as the deviation of measured value from the true value. On the
other hand, the precision of measurement is defined as the deviation of different readings
from the mean value. Thus, it is measure of consistency in measurement. An example
will clarify this point. The value of a known voltage source of 100 V source is measured
with a voltmeter. Five readings were taken. The indicated readings were 103 V, 105 V,
104 V, 103 V, 105 V. In this case, the accuracy of the instruments is better than 5%,
because the maximum deviation from true value is 5 V. But the precision of the
instrument is + 1 V because the deviation of the readings from mean value is + 1 V.
SAQ 2
(a) What is the difference between accuracy and precision?
(b) What do you mean by the ‘accuracy of the instrument is better than 2%’?

2.4 CALIBRATION OF THE INSTRUMENT

The calibration of the instrument is done to find its accuracy. Before using an instrument,
particularly a new one, in a measurement system, it is required to calibrate it to find the
accuracy, precision or uncertainty of the instrument. It can be done by comparing its
performance with
(a) a primary standard instrument,
(b) a secondary standard instrument having high accuracy, and
(c) a known input source.
For example, a flowmeter might be calibrated by
(a) comparing it with a standard flow measurement facility of the National
Bureau of Standards,
(b) comparing it with another flowmeter of known accuracy, or
(c) direct calibration with a primary measurements such as weighing a certain
amount of water in a tank and recording the time elapased for the quantity of
flow through the meter.
22
SAQ 3 Errors in Measurement

(a) What is the need of calibration of a measuring instrument?

(b) Mention the procedures of calibrating a pressure gage.

2.5 ANALYSIS OF THE ERRORS

When an experiment is performed and some data are obtained, then it is required to
analyse these data to determine the error, precision and general validity of the
experimental measurements. Bad data due to obvious blunder or reason may be discarded
immediately. We cannot throw out the data because they do not conform with our hopes
and expectations unless we see something obviously wrong. If such bad points fall
outside the range of normally expected random deviations, they may be discarded on the
basis of some consistent statistical analysis. The elimination of data point must be
consistent and should not be dependent on human whims and biased based on what
‘ought to be’. In many instances, it is very difficult for the individual to be consistent and
unbiased. The presence of a deadline, disgust with previous experimental failures, and
normal impatience all can influence rational thanking processes. However, the competent
experimentalist will strive to maintain consistency in the primary data analysis.
2.5.1 Error Analysis on Common Sense Basis
Analysis of the data on common sense basis has many forms. One rule of thumb that
could be used is that the error in the result is equal to maximum error in any parameter
used to calculate the result. Another commonsense analysis would combine all the errors
in the most detrimental way in order to determine the maximum error in the final result.
Consider the calculation of electric power from
P=E.I . . . (2.1)
where, E and I are measured as :
E = 100 V ± 2 V . . . (2.2)
I = 10 A ± 0.2 A . . . (2.3)
The nominal value of power is 100 × 10 = 1000 W. By taking worst possible variations
in voltage and current, we could calculate
Pmax = (100 + 2) (10 + 0.2)
= 1040.4 W
Pmin = (100 – 2) (10 – 0.2)
= 960.4 W
Thus, using the method of calculation, the uncertainty in the power is + 4.04 percent or
– 3.96 percent. It is quite unlikely that the power would be in error by these amounts
because the voltmeter variation would probably not correspond with the ammeter
variations. When voltmeter reads an extreme ‘high’, there is no reason why the ammeter
must also read an extreme ‘high’ at that particular instant, indeed, this combination is
most unlikely.
The simple calculation applied to the electric-power equation above is a useful way of
inspecting experimental data to determine what error could result in a final calculations.
However, the test is too severe and should be used only for rough inspection of data. It is
significant to note, however, that if the results of the experiments appear to be in error by
more than the amounts indicated by the above calculation, then the experimenter had
better examine the data more closely. In particular, the experimenter should look for
certain fixed errors in the instrumentation, which may be eliminated by applying either
theoretical or empirical corrections.
23
Principle of As another example, we might conduct an experiment where heat is added to a container
Measurementgs and of water. If our temperature instrumentation would indicate a drop in a temperature of the
Measuring Techniques
water, our good sense would tell us that something is wrong and the data point(s) should
be thrown out. No sophisticated analysis procedures are necessary to discover this kind
of error. The term ‘common sense’ has many connotations and means different things to
different people. In the brief example given above, it is intended as a quick and expedient
vehicle, which may be used to examine experimental data and results for gross errors and
variations.
2.5.2 Uncertainty Analysis
We have already discussed that it is more explicit to speak of experimental uncertainty
rather than experimental error. Uncertainty analysis is an useful to examine experimental
data. The method is based on a careful specifications of the uncertainties in the various
primary experimental measurements. For example, a certain pressure reading might be
expressed as
P = 100 kN/m2 ± 1 kN/m2
When the plus or minus notation is used to designate the uncertainty, the person making
this designation is stating the degree of accuracy with which he or she believes the
measurement has been made. We may note that this specifications is in itself uncertain
because the experimenter is naturally uncertain about the accuracy of these
measurements.
If a very careful calibration of an instrument has been performed recently, with standard
of very high precision, then the experimentalist will be justified in assigning a much
lower uncertainty to measurements than if they were performed with a gage of instrument
of unknown calibration history.
To add further specification of the uncertainty of a particular measurement, two scientists
Kline and Meclintoek propose that the experimenter specify certain odds for the
uncertainty. The above equation for pressure might thus be written as
P = 100 kN/m2 ± 1 kN/m2 (20 to 1)
In other words, the experimenter is willing to bet with 20 to 1 odds that the pressure
measurement is within ± 1 kN/m2. It is important to note that the specification of such
odds can only be made by the experimenter based on the total laboratory experience.
Suppose, a set of measurements is made and the uncertainty in each measurement may be
expressed with the same odds. These measurements are then used to calculate some
desired results of the experiments. We wish to estimate the uncertainty in the calculated
result on the basis of the uncertainties in the primary measurements. The result R is a
given function of the independent variables x1, x2, x3 , . . . , xn. Thus
R = R (x1, x2, x3, . . . , xn) . . . (2.4)
Let WR be uncertainty in the result and W1, W2, . . . , Wn be the uncertainties in the
independent variables. If the uncertainties in the independent variables are all given with
same odds, then the uncertainty in the result having these odds is given as
1/ 2
⎡⎛ ∂ R ⎞
2
⎛ ∂R ⎞
2
⎛ ∂R ⎞ ⎤
2
⎢
WR = ⎜ W1 ⎟ + ⎜ W2 ⎟ + . . . + ⎜ Wn ⎟ ⎥ . . . (2.5)
⎢⎝ ∂ x1 ⎠ ⎝ ∂ x2 ⎠ ⎝ ∂ x ⎠ ⎥⎦
⎣ n

Example 2.1
The resistance of a certain size of wire is given as
R = R0 [1 + α (T − 20)]
where, Ro = 6 Ω ± 0.3 percent is the resistance at 20oC,
α = 0.004oC–1 ± 1 percent is the temperature coefficient of resistance, and
T = 30 ± 1oC is the temperature of the wire.
Find its uncertainty.
24
Solution Errors in Measurement

The nominal resistance is

R = (6) [1 + (0.004) (30 – 20)]
= 6.24 ohms
The uncertainty in this value can be obtained by applying Eq. (2.5)
The various terms are
∂R
= 1 + α (T − 20) = 1 (0.004) (30 − 20) = 1.04
∂R0
∂R
= R0 (T − 20) = 6 (30 − 20) = 60
∂α
∂R
= R0 α = 6 × 0.004 = 0.024
∂T
WR0 = 6 × 0.003 = 0.018 ohm

Wα = 0.004 × 0.01 = 4 × 10−3 o C −1

WT = 1o C
Thus, the uncertainty in the resistance is
WR = [(1.04) 2 (0.018) 2 + (60) 2 (4 × 10− 3 ) 2 + (0.024) 2 (1) 2 ]1/ 2
= 0.0305 ohm
0.0305
% WR = × 100 = 0.49%
6.24
2.5.3 Statistical Analysis of Experimental Data
It is important to define some pertinent terms before discussing some important methods
of statistical analysis of experimental data.
Arithmetic Mean
When a set of readings of an instrument is taken, the individual readings will vary
somewhat from each other, and the experimenter is usually concerned with the
mean of all the readings. If each reading is denoted by xi and there are n readings,
the arithmetic mean is given by
n
1
xm =
n
∑ xi . . . (2.6)
i =1

Deviation
The deviation, d, for each reading is given by
di = xi − xm . . . (2.7)
We may note that the average of the deviations of all readings is zero since
n n
1 1
di =
n
∑ di = n ∑ ( x i − x m )
i =1 i =1

1
= xm − (nx m )
n
=0 . . . (2.8)
The average of the absolute value of the deviations is given by
n
1
|d i | =
n
∑ | di |
i =1
25
Principle of 1 n
Measurementgs and
Measuring Techniques
=
n
∑ [ xi − x m ] . . . (2.9)
i =1

Note that the quantity is not necessarily zero.

Standard Deviation
It is also called root mean-square deviation. It is defined as
1/ 2
⎡1 n ⎤
σ=⎢ ∑ ( xi − x m ) ⎥ 2
. . . (2.10)
⎢⎣ n i =1 ⎥⎦
Variance
The square of standard deviation is called variance. This is sometimes called the
population or biased standard deviation because it strictly applies only when a
large number of samples is taken to describe the population.
Geometrical mean
It is appropriate to use a geometrical mean when studying phenomena which grow
in proportion to their size. This would apply to certain biological processes and
growth rate in financial resources. The geometrical mean is defined by
1
xg = [ x1 . x2 . x3 . . . xn ]n . . . (2.11)
Example 2.2
The following readings are taken of a certain physical length. Compute the mean
reading, standard deviation, variance and average of the absolute value of the
deviation using the biased bases.
Reading 1 2 3 4 5 6 7 8 9 10
xi (cm) 5.30 5.73 6.77 5.26 4.33 5.45 6.09 5.64 5.81 5.75

Solution
n
1 1
xm =
n
∑ xi =
10
(56.13)
i =1

= 5.613 cm
The other quantities are computed with the aid of the following table.
Reading di = xi – xm (xi – xm)2
1 − 0.313 0.09797
2 0.117 0.01369
3 1.157 1.33865
4 − 0.353 0.12461
5 − 1.283 16.4609
6 − 0.163 0.02657
7 0.477 0.22753
8 0.027 0.00729
9 0.197 0.03881
10 1.137 0.01877
1/ 2
⎡1 n ⎤
σ = ⎢ ∑ ( x i − x m )2 ⎥
⎣ n i =1 ⎦

26
 1/ 2 Errors in Measurement
⎡1 ⎤
= ⎢ (3.533) ⎥ = 0.5944 cm
⎣10 ⎦

σ 2 = 0.3533 cm 2

1 n 1
|d i | = ∑ | di | = ∑ | xi − xm |
n i =1 n
1
= × 4.224 = 0.4224 cm
10
Example 2.3
The 5 years record of a mutual fund investment is shown below. Find the average
growth rate.

Year Asset Value Rate of Increase Over

Previous Year
1 1000 −
2 890 0.89
3 990 1.1124
4 1100 1.1111
5 1290 1.1364

Solution
Average growth rate = [(0.89) (1.1124) (1.1111) (1.1364)]1/4
= 1.0574
To see that this is indeed a valid average growth rate we can observe that
(1000) (1.0574)4 = 1250
2.5.4 Method of Least Squares
Suppose, we have a set of observations x1, x2, . . . , xn. The sum of the squares of their
deviations from some mean value is
n
S = ∑ ( xi − xm )2 . . . (2.12)
i =1

Now, suppose, we wish to minimize S with respect to the mean value. We set
∂S n
= 0 = ∑ − 2 ( xi − xm )
∂xm i =1

⎛ n ⎞
= − 2 ⎜ ∑ xi − n xm ⎟ . . . (2.13)
⎝ i =1 ⎠
where, n is the number of observations.
We find that
1 n
xm = ∑ xi . . . (2.14)
n i =1
or mean value that minimizes the sum of the squares of deviations is the arithmetic mean.
This example might be called the simplest application of the method of least square. We
shall be able to give only two other applications of the method, but it is of great utility in
analyzing experimental data.
Suppose that the two variables x and y are measured over a range of values. Suppose,
further that we wish to obtain a simple analytical expression for y as a function of x. The 27
Principle of simplest type of function is a linear one, hence we might try to establish y as a linear
Measurementgs and function of x. Both x and y may be complicated functions of other parameters so arranged
Measuring Techniques
that x and y vary approximately in a linear manner. The problem is one of finding the
best linear function, for the data may scatter a considerable amount. We could solve the
problem rather quickly by plotting the data points on graph paper and drawing a straight
line through them by eye.
Indeed this is common practice, but the method of least squares gives a more reliable way
to obtain a better functional relationship than the guesswork of plotting. We seek an
equation of the form.
y = ax + b . . . (2.15)
We, therefore, wish to minimize the quantity
n
S = ∑ [ yi − (a xi + b)]2 . . . (2.16)
i =1

This is accomplished by setting the derivatives with respect to ‘a’ and ‘b’ zero.
Performing these operation, there results
∑ yi = n b + a ∑ xi . . . (2.17)

∑ xi yi = b ∑ xi + a ∑ xi2 . . . (2.18)
Solving Eqs. (2.17) and (2.18)
n ∑ xi yi − (∑ xi ) (∑ yi )
a= . . . (2.19)
n ∑ xi2 − (∑ xi )2

(∑ yi ) (∑ xi2 ) (∑ xi yi ) (∑ xi )
b= . . . (2.20)
n ∑ xi2 − (∑ xi ) 2
Designating the computed value of
y = ax + b . . . (2.21)

And the standard error of estimate of y for the data is

1/ 2
⎡ ∑ ( yi − yi ) 2 ⎤
Standard Error = ⎢ ⎥
⎣⎢ n−2 ⎦⎥
1/ 2
⎡ ∑ ( yi − a xi − b)2 ⎤
=⎢ ⎥ . . . (2.22)
⎢⎣ n−2 ⎥⎦
The method of least squares may also be used for determining higher-order polynomials
for fitting data. One only needs to perform additional differentiations to determine
additional constant. For example, it is desired to obtain a least-squares fit according to
the quadratic function
y = ax2 + bx + c . . . (2.23)
n
S = ∑ [ yi − (axi2 + bxi + c)]2 . . . (2.24)
i =1

Would be minimized by setting the following derivatives equal to zero.

∂S
= 0 = ∑ 2 [ yi − (axi2 + bxi + c) (− xi2 )]
∂a
∂S
= 0 = ∑ 2 [ yi − (axi2 + bxi + c) (− xi )]
∂b

28
 ∂S Errors in Measurement
= 0 = ∑ 2 [ yi − (axi2 + bxi + c) (− 1)]
∂c
Expanding and collecting terms
a ∑ xi4 + b ∑ xi3 + c ∑ xi2 = ∑ xi2 yi . . . (2.25)

a ∑ xi3 + b ∑ xi2 + c ∑ xi = ∑ xi yi . . . (2.26)

a ∑ xi2 + b ∑ xi + n c = ∑ yi . . . (2.27)
These equations may then be solved for constants a, b and c.
Example 2.4
Following data points are expected to follow a functional relation in the form of
y = ax + b. Find a best functional relation between x and y using the method of
least squares.
xi 1.0 1.6 3.4 4.0 5.2
yi 1.2 2.0 2.4 3.5 3.5

Solution
xi yi xi2 xi yi n
1.0 1.2 1.0 1.2 5
1.6 2.0 2.56 3.2
3.4 2.4 11.56 8.16
4.0 3.5 16.0 14.0
5.2 3.5 27.04 18.2
∑ xi = 15.2 ∑ yi = 12.6 ∑ xi2 = 58.16 ∑ xi yi = 44.76

We calculate the value of a and b using Eqs. (2.19) and (2.20)

(5 × 44.76) − (15.2) (12.6)
a=
(5 × 58.16) − (15.2)2
= 0.54
(12.6) × 58.16 − (44.76 × 15.2)
b=
(5 × 58.16) − (15.2) 2
= 0.879
Thus, the desired relation is
y = 0.540 x + 0.879

2.5.5 Analysis of the Probability of Errors

The probability of a random variable is defined as the number of occurrence of the
variable (in per unit of the number of experiments) if the infinite number of experiments
are repeated.
The probability of a random variable X, equal to a number a, i.e., X = a, is denoted by
P (X = a). Similarly, probability of the random variable X, assuming any value in the
interval (a, b), i.e. a < X < b, is denoted by P (a < X < b), if – α < X < α, the probability is
denoted by P (−α < X < α) which is equal to unity, i.e.
P = (− α < X < α ) = 1 . . . (2.28)
Let x1, x2, . . . , xn (n being large) be the values for which random variable x has a positive
probability and let the corresponding probabilities be P1, P2, . . . , Pn.
Now, let us introduce a function
29
Principle of ⎧ p , when x = x i (i = 1, 2, . . . , n)
Measurementgs and f ( x) = ⎨ i . . . (2.29)
Measuring Techniques ⎩ 0 otherwise
f (x) is called probability function (descrete probability density function) of x.
The probability P (X ≤ x) depends on the value of x and is a function of x, which is called
the probability distribution function of x and is denoted as
F ( x) = P ( X ≤ x) . . . (2.30)
In case x being discrete variable, the probability distribution function is given by
F ( x ) = ∑ f ( xi ) . . . (2.31)
xi ≤ x

If the random variable x is continuous, then

x
F ( x) = ∫ f ( y ) dy . . . (2.32)
−α

A very important type of probability distribution known as Gaussian or Normal

distribution. It is a continuous probability distribution with probability density, given as
2
1 ⎡x − x ⎤
− ⎢
1 σ ⎥⎦
f ( x) = e 2⎣ . . . (2.33)
σ 2π
A random variable with this probability distribution is said to be normal or normally
distributed. The Gaussian probability distribution is very important because many
random variables of practical interest are normal or approximately normal or can be
transformed into normal random variables in a relatively simple way. Three curves of
f (x) for different values of σ are shown in Figure 2.2.
f (x)

σ = 0.25
1

σ = 0.5

σ = 1.0

X–2 X–1 X X+1 X+2

Figure 2.2 : Gaussian Distribution Curve

The mathematical treatment of Gaussian law is beyond the scope of this course, so we
shall give the following qualitative statements of the Gaussian distribution.
(a) The sample data with small deviations are more probable than the data with
large deviations.
(b) Data with large deviations are very less probable.
(c) There is an equal probability of plus and minus deviations so that the curve
is symmetrical about the mean value.
(d) About 50% of the data will lie between x – 0.67450 σ and x + 0.6745 σ.
30 (e) About 68% of the data will lie between x – σ and x + σ.
 (f) About 95% of the data will lie between x – 2 σ and x + 2 σ. Errors in Measurement

(g) About 99% of the data will lie between x – 3 σ and x + 3 σ.

Probable Error
It has been mentioned that the 50% of the observation have been the deviation
limits of ± 0.6745 σ. The probable error is defined as this limiting deviation, i.e.
Probable Error = ± 0.6745 σ
2.5.6 Analysis of Limiting Errors
Manufacturers of any equipment give guarantee about the accuracy of the equipment
with some limiting deviations from the specified accuracy. Most of the measuring
instruments are guaranteed for their accuracy with a percentage deviation of full scale
reading. This limiting deviation from the specified values are called the limiting error or
guarantee error. For example, if the pressure of pressure gage is specified as 100 kN/m2 ±
1%, the pressure of the gage may have any value between 90 kN/m2 to 110 kN/m2.
Combination of Limiting Errors
If quantities with limiting errors are combined together according to some
algebraic laws, the limiting error in the resulted quantity may be calculated as
follows :
Sum or Difference of Quantities
Let y = u ± v ± w and Δu, Δv and Δw be errors in u, v and w respectively. If
Δy is the error in y, then
y + Δy = (u + Δu ) ± (v + Δv) ± ( w + Δw)
= (u ± v ± w) ± (Δu ± Δv ± Δw)
or, Δy = Δu ± Δv ± Δw

Δy Δu ± Δv ± Δw
or, =
y y

u Δu v Δv w Δw
= . ± . ± .
y u y v y w

u
∴ % Error in y = × % error in u
y

v
± × % error in v
y

w
± × % error in w
y
Similarly, limiting errors for product of the quantities and quotient of the quantities can
be obtained.
SAQ 4
(a) What is the need of analysis of an experimental data?
(b) What is the difference between error and uncertainty?
(c) What do you mean by limiting error?
(d) Following data points are expected to follow a functional variation between
x and y in the form of

y = a eb x 31
Principle of Find the best functional relation between x and y using the method of least
Measurementgs and squares.
Measuring Techniques
x 1 2 3 4 5
y 8.0 7.2 6.5 4.2 2.5

(e) Three elements have following ratings :

R1 = 40 ± 5%, R2 = 80 ± 5%, R3 = 50 ± 5%
where, Rs = R1 + R2 + R3. Find the magnitude of Rs and the limiting errors in
Rs and in percentage of three elements.
(f) The length and width of a rectangular plate are (0.163 ± 0.0005) m and
(0.138 ± 0.0005) m respectively. Calculate the area of the plate and also
uncertainity in the area.
(g) A physical quantity P is related to four parameters a, b, c and d as follows :
a3 b 2
P = 1/ 2 . The percentage error of measurement in a, b, c and d are 1%,
c
3%, 4% and 2% respectively. What is the percentage error in the
quantity P?

2.6 SUMMARY
Error in the measurement of a physical quantity indicates the deviation from its actual
value.
Errors can be classified as Gross error, Systematic error and Random error.
Accuracy and precision play important roles in the measurement of any physical
quantity. Calibration of an instrument is done to find its accuracy. It can be done either
by
(a) comparing with a standard instrument,
(b) comparing with an another instrument with known accuracy, or
(c) direct calibration with primary measurement.
When an experiment is performed and some data are obtained, then it is required to
analyse these data to find error, precision and the general validity of the experimental
measurements.
The error analysis of the experimental data can be done by various methods, such as
common sense basis, uncertainty analysis, statistical analysis, probability error analysis,
limiting error analysis etc.

2.7 KEY WORDS

Gross Errors : These are due to the gross blunder on the part of
the experimenters or observers.
Systematic Errors : These are inherent errors of apparatus or method.
32
Random Errors : Their magnitudes are not constant. The law of Errors in Measurement
probability may be applied to these errors.
Accuracy : Deviation of the measured value from true value.
Precision : Deviation of the different readings from mean
reading.
Calibration : To make a comparison to find the accuracy.
The Method of Least Squares : Mean value that minimizes the sum of the squares
of the deviations is the arithmetic mean.

2.8 ANSWERS TO SAQs

SAQ 1
(a) Error, in the measurement of a physical quantity, is the deviation from its
actual value, whereas the precision is the deviation of some readings from
their mean value.
(b) Experimental errors cause uncertainty in the results of the experimental
measurements.
(c) Fixed error always gives constant deviation and it is one kind of systematic
error. The random error is caused by personal fluctuations, random
electronic fluctuation in the apparatus or instruments etc.
(d) The experimenter can minimise the error in the measurement by knowing
the reasons of the error and adopting correct procedures for the experiments.
(e) (i) Random error
(ii) Systematic error
(iii) Random error
(iv) Gross error
SAQ 2
(a) The accuracy is the deviation of a measured value from its true value,
whereas precision is the deviations of some readings from their mean value.
(b) The maximum possible error in the instrument is 2% and it is 98% accurate.
SAQ 3
(a) The calibration of an instrument is required to find its accuracy.
(b) A pressure gage can be calibrated
(i) By comparing with a standard pressure gage (ISI standard).
(ii) By comparing with another pressure gage with known accuracy.
(iii) By direct measurement of pressure against a known value.
SAQ 4
(a) When an experiment is performed and some data are obtained, then it is
necessary to analyse these data to find error, precision and the general
validity of the experimental measurements.
(b) The error is the deviation of the measured value from the true value,
whereas uncertainty denotes the possible value the error may have, and it
may vary in great deal depending upon the circumstances of the experiment.
(c) Most of the measuring instruments are guaranteed for their accuracy with a
percentage deviation of full scale reading. This limiting deviation from the
specified values are called limiting errors.
(d) y = aeb x
⇒ ln y = ln a + bx
33
Principle of ⇒ Y = AX + B
Measurementgs and
Measuring Techniques where, Y = ln y
X =x
A=b
B = ln a

xi yi Xi Yi Xi Yi Xi2 n
(= xi) (= In yi)
1 8.0 1 2.079 2.097 1 5
2 7.2 2 1.974 3.948 4
3. 6.5 3 1.872 5.616 9
4. 4.2 4 1.435 5.74 16
5. 2.5 5 0.976 4.875 25
∑ X i = 15 ∑ Yi = 8.276 ∑ X i Yi = 22.258 ∑ X i2 = 55

n ∑ X i Yi − (∑ X i ) (∑ Yi )
Now, A=
n (∑ X i2 ) − (∑ X i )2
5 × 22.258 − 15 × 8.276
=
5 × 55 − 152
= − 0.257
=b
(∑ Yi ) (∑ X i2 ) − (∑ X i Yi ) (∑ X i )
B=
n (∑ X i2 ) − (∑ X i ) 2
8.276 × 55 − 22.258 × 15
=
50
455.18 − 333.87
=
50
= 2.426
= ln a
∴ a = 11.313
Hence, the best functional relation between y and x is
y = 11.313 e − 0.257 x
(e) RS = R1 + R2 + R3
= 40 + 80 + 50
= 170
R1 ΔR1 R2 ΔR2 R3 ΔR3
% error in RS = + +
RS R1 RS R2 RS R3
40 80 50
= × 5% + × 5% + × 5%
170 170 170
5%
= [40 + 80 + 50]
170
= 5%
5
34 % error in RS = × 170 = 8.5
100
(f) Length = (0.163 ± 0.0005) m; Δ l = 0.0005 m Errors in Measurement

Width = (0.138 ± 0.0005) m; Δ b = 0.0005 m

Δl 0.0005
∴ = = 0.003
l 0.163
Δb 0.0005
= = 0.0036
b 0.138
ΔA Δl Δb
= + = 0.003 + 0.0036 = 0.0066
A l b
Relative % error = 0.0066 × 100 = 0.66%
Nominal area = 0.163 × 0.138 m2 = 0.0225 m2
Uncertainity in the area = (0.0066 × 0.0255) m2
Area of the plate = (0.0225 ± 0.00015) m2
Δρ
(g) Maximum possible error in ρ = × 100
ρ
Δρ Δa Δb
× 100 = 3 × 100 + 2 × 100
ρ a b
1 Δc Δd
+ × 100 + × 100
2 c d
1
= 3 × 1% + 2 × 3% + × 4% + 1 × 2%
2
= (3 + 6 + 2 + 23) %
= 13%

35